Intonation in Déise Irish

ENGAGING WITH ROBUST CROSS-PARTICIPANT VARIABILITY IN AN ENDANGERED MINORITY VARIETY: INTONATION IN DÉISE IRISH Connor McCabe Trinity College, Dublin [email protected] ABSTRACT This paper adapts data and analyses from unpublished recent work on Déise prosody [11], This paper explores the issue of speaker uniformity in elaborating on the importance of engaging with and the phonetic study of endangered language varieties, seeking to explain participant variability. Distribution with reference to work on the dialect of Irish (Gaelic) of intonational features (described autosegmentally- spoken in Gaeltacht na nDéise (Co. Waterford). metrically using IViE; [7]) is compared with Detailed prosodic study of this subvariety of Munster participant age and relative ‘traditionalness’. A 10- Irish directly engaged with variation across point traditionalness scale based on phonological, generations and degrees of ‘traditionalness’. Age and lexical, morphosyntactic, and acquisition factors is score on a 10-point traditionalness scale showed no used, experimenting with a quantitative approach to correlation with one another, justifying the intuitions found in the literature on variation in consideration of the two as distinct factors. endangered speaker populations [5,8,10,14]. A falling H*+L predominated in both prenuclear and nuclear position. Relative distribution of pitch Figure 1. Map of Ireland with Gaeltacht areas indicated accent types (H*+L, H*, L*+H) and boundary tones in bold [16], and arrow indicating the Déise (H%, 0%) frequently correlated with participant age, and more rarely with traditionalness score. The importance of a realistic view of interspeaker variability in endangered varieties, and how to approach this quantitatively, is discussed. Keywords: Intonation, variation, sociophonetics, endangered varieties 1. INTRODUCTION . A degree of interspeaker variation is expected in the 2. BACKGROUND study of any language. The numerous social and linguistic pressures affecting endangered varieties 2.1 Participant variation may cause this variation to be more pronounced [15]. Irish (Gaelic), though officially the primary The primary goal of [11] was to create a pitch accent language of the Republic of Ireland, is spoken and boundary tone inventory that would be natively by only a small percentage of the population. comparable with existing data for other dialects of Historically continuous Irish-speaking populations Irish [4,6]. Interviews were conducted using a Zoom are scattered around the periphery of the island in H4n recording device with eleven participants who all regions known as Gaeltachtaí. This paper focusses on identified as native speakers of the local variety. Wide the microdialect of Irish spoken in the Waterford idiolectal variation was immediately evident. Gaeltacht, also known as Gaeltacht na nDéise. Déise Participants were asked a series of biographical Irish is a subvariety of the Munster macrodialect, and questions (e.g. what language each parent spoke, has a number of distinct phonological, lexical, and which language predominated in the home, morphosyntactic features. Like other dialects, Déise language(s) of education, etc.), and a continuum Irish is under considerable pressure from English. emerged between more and less traditional speakers. However, its extreme subminority status (1.7% of the The former showed fewer signs of national Gaeltacht population; [16]) means that it is supraregionalisation of their idiolect; the latter also subject to influence from dialects of Irish more included both those with mixed acquisition situations robustly represented in Irish-language media, and those with significant influence from other especially those of Kerry and Conamara. dialects of the language. The unpredictable nature of acquisition in endangerment situations, especially when a gradiently similar macrovariety is available Irish were able to include areal microvariation [4,6]. (see [10] on Louisiana Creole, Louisiana French, and Munster Irish was represented only by Kerry [4], the Standard French) interrupts the expected link most robust subdialect. The only existing description between speaker age and relative conservatism of intonation in the other Munster subdialects of Cork [5,8,10]. and the Déise are anecdotal descriptions in In order to operationalise measurement of the dialectological works from the 1940s [3,13]. ‘traditionalness’ intuition, a rough 10-point scale was Dalton [4] found an overwhelming preference for devised. Simple features for which presence/absence falling pitch accents (H*+L) in Kerry Irish. Falls was easily identifiable from interviews were selected. comprised 100% of nuclear accents in all sentence Half points were included when participants types (declarative, Yes/No-questions, and WH- demonstrated awareness of a feature, but failed to use questions). There was a small minority of prenuclear it in their own speech. Four segmental features were highs (H*) and rises (L*+H), the latter always selected, along with two lexical features, one prosodic preceded by a high boundary tone (%H). [2] and [13] feature, two morphological features, and one describe (impressionistic) nuclear rises and highs in ‘acquisition’ feature (with half a point awarded per the Déise and Cork. Irish-speaking parent). A full description of the It was expected that Déise intonation would features selected can be found in Appendix A of [11]. roughly pattern with Kerry Irish. The possibility of Participant age showed no correlation with nuclear rises and/or highs was borne in mind. traditionalness score (r2=0.0214), supporting the independence of the two factors. 2.3 Elicitation Table 1. Lowest, highest, and median scores from Participants were recorded in a quiet room in a local [10]’s 11 participants. Age included in brackets below school. In order to ensure maximum comparability of participant label. data, the same corpus of declarative sentence, Y/N- Feature1 06 02 NC2 questions, and WH-questions used in [4] was adapted (56) (49) (94) for use. The sentence list totalled 67 items, and 3 Diphthongisation 1 1 1 repetitions were attempted for all participants. In e.g. Rinn /raɪnj/ some cases, this was cut short for reasons of stamina <-th> as [x] 0 .5 1 or time constraint. A short story, Bean an Leasa, was also read, consistent with methodology in other IViE /lγ/ as [γ] 0 0 .5 work [7]. -/N′/# as [ŋ] 0 1 1 LEXIS: bleán>crú .5 .5 1 2.4 Analysis ‘milk.VERB’ LEXIS: clois>airigh 0 0 1 Sentence list and short story readings were analysed ‘hear’ in Praat [1], using IViE conventions. The decision to Past-tense 0 .5 1 use IViE was in keeping with the aim of marker <Dh’-> comparability with previous work on Irish. The IViE Choˈnaic ‘saw’ 0 1 1 provision for a neutral boundary tone option (0%) is (Final stress) particularly useful for the parsimonious analysis of Irish data [4,6]. Levels of analysis were gradually sa + eclipsis 0 .5 1 layered onto one another, allowing for frequent Irish-speaking 1 .5 1 checks of analytical consistency across files. parents Once analysed, distribution of intonational Score: 2.5 5.5 9.5 features (as a percentage of all cases) was compared with participant age and traditionalness score. 2.2 Previous work on Irish intonation Potential correlations were evaluated using the coefficient of determination r2, with a consideration Work as part of the Prosody of Irish Dialects (PoID) threshold of r2≥0.2 based on psychological literature project [12], and subsequent related research, [3]. The use of a descriptive rather than inferential provides information on pitch accents, boundary statistic was appropriate to both the exploratory tones, and alignment patterns in the three Irish nature of the work and the sample size in question [9]. macrodialects. Investigations of Connacht and Ulster 1 For explanation of these feature labels, see chapters 3.3 2 NC did not receive a numerical label, as his interview and 5.1 of [11], available online at https://bit.ly/2BRbcZo was unique. He was unable to complete the standard elicitation task due to his age and health. 3. RESULTS 3.2 Nuclear Pitch Accents Findings for pitch accent and boundary tone Nuclear pitch accent distribution roughly parallels inventory and distribution are summarised below. that of Kerry. However, while H*+L is the only Results for statistical comparison of age and nuclear pitch accent attested in the latter variety, traditionalness as predictors of individuals’ Déise participants exhibited notable variation. H*+L distributions are reported; the issue of evident was consistently the dominant nuclear pitch accent, participant variability is discussed in section 4. but only for a single participant was it the exclusive nuclear accent type attested. Distribution of nuclear 3.1 Prenuclear Pitch Accents pitch accents for all sentence types showed some degree of correlation with participant age. Distribution of prenuclear pitch accents in the Déise In declaratives, participant age emerged as a closely resembled that described by [4] for Kerry predictor for (i) relative strength of H*+L Irish. There was a strong preference for falling H*+L (r2=0.2379), and (ii) relative weakness of L*+H (76-97% of all prenuclear accents), with a notable (r2=0.37). This is consistent with the youngest minority of prenuclear rises (L*+H; 1-17%), and a participant (10; 34 years old) exhibiting nuclear rises more scarce H* (0-10%). in 27% of her declaratives, versus 9% of declaratives Findings for the three sentence types elicited are for the two participants with the next highest L*+H considered in sequence below. usage. This result compliments 10’s use of nuclear Distribution of prenuclear pitch accents in H*+L H%, which also highlights a degree of declaratives were best predicted by participant age. ambiguity in assignment of accent labels. 2 This applied to H*+L (r =0.2055) and L*+H For WH-questions, the only significant correlation 2 (r =0.29), but not to H* which showed relationships to emerge was between degree of nuclear L*+H with neither age nor traditionalness. Older speakers presence and participant age (r2=0.2035).

Load more